Role of Strong Zeolitic Acid Sites on Hydrocarbon Reactions

Mar 22, 2008 - Baodong Wang and George Manos*. Department of ... can be attributed to the strong zeolite acidity, and the reaction can be described we...
0 downloads 0 Views 193KB Size
2948

Ind. Eng. Chem. Res. 2008, 47, 2948-2955

Role of Strong Zeolitic Acid Sites on Hydrocarbon Reactions Baodong Wang and George Manos* Department of Chemical Engineering, UniVersity College London, Torrington Place, London WC1E 7JE, UK

In the work reported in this article, 1-pentene reactions were carried out over USHY zeolite, whose strongacid sites were selectively poisoned by hard coke to study the role of these strong zeolitic acid sites on hydrocarbon conversions. We show conclusively that strong-acid sites are responsible for cracking and hydride transfer reactions as well as strong coke formation, whereas weak-acid sites can only catalyze double-bond isomerization. 1. Introduction The performance of zeolite catalysts in hydrocarbon reactions depends on their acidity, in particular number and strength of the acid sites.1 It is generally believed that the cracking activity can be attributed to the strong zeolite acidity, and the reaction can be described well with carbenium ion chemistry.2 In this article, we look at the role of strong acidity in hydrocarbon reactions. For that, we selectively poisoned the strong-acid sites of an ultrastable Y zeolite (USHY) and tested the resulting catalyst samples in the same reaction as the fresh catalyst. For this selective poisoning, we used coke components and specifically hard coke. During organic reactions catalyzed by solid acidic catalysts, the catalyst always suffers from strong deactivation due to formation and retention of heavy byproducts, so-called coke, which deactivate acid sites.3 Coke components can be classified into coke precursors and hard coke.4 Coke precursors are the coke components that can be removed from the coked catalyst sample simply through volatilization in inert nitrogen, whereas hard coke remains on the catalyst even at high temperature (873 K) and is removed by burning. The relation between acidity-concentration as well as strength and activity has been long sought for catalytic reactions. Kung and co-workers set up a model to study the roles of acid strength in cracking activity.5 The understanding in depth of the relation between acidity and activity allows the further study of the catalyst deactivation as well as the development of improved catalysts, generating less coke and being less sensitive to deactivation. The objective of the present work was to study the role of strong-acid sites on hydrocarbon reactions by selectively poisoning strong zeolitic acid sites with hard coke. The dependence of product distribution, reactant conversion, and catalyst deactivation upon catalyst acidity strength was determined during a 1-pentene reaction over different catalyst samples, fresh and pre-coked, at various conditions. Moreover, acid site characterization of the used catalysts was also investigated. 2. Experimental Section 2.1. Materials. The catalyst used was an ultrastable Y zeolite in acidic form (USHY) provided by Grace Gmbh with a framework Si/Al ratio of 5.7 and an overall Si/Al ratio of 2.5. The micropore area was 532.4 m2/g, and the micropore volume was 0.26 cm3/g. The measured N2 BET surface area was 590 * To whom correspondence should be addressed. Tel.: +44 (0)20 7679 3810. Fax: +44 (0)20 7383 2348. E-mail: [email protected].

( 23.5 m2/g. The hydrocarbon reactant, 1-pentene (99% purity), was supplied by Sigma-Aldrich Chemicals. Carrier gas Nitrogen (CP grade) was supplied by the BOC group. 2.2. Coke Classification and Acidity Characterization. Coke classification was carried out as described in previous work using a thermal gravimetric apparatus (TGA), Cahn TG 131.4,6 About 150 mg coked catalyst was heated from room temperature to 473 K with 10 K/min and remained there for 60 min under nitrogen flow (60 mLN/min) to remove adsorbed water and reaction-mixture components. Then, the temperature was increased to 873 K with 10 K/min and stayed at 873 K for 30 min under the same nitrogen flow. During this period, coke precursors were removed, resulting in a sample-weight decrease. By switching from nitrogen to air at the final temperature (873 K) and at the same flow rate, the hard coke deposited on the catalyst was burnt off and its weight was measured. The amount of coke precursors in the catalyst was calculated as the difference between the sample mass after drying at 473 K and just before switching from nitrogen to air at 873 K. The amount of hard coke was estimated by the mass difference of the catalyst sample between before and after switching from nitrogen to air, when the hard coke was completely burnt off. All of the coke content, expressed in percentage, was estimated by dividing the corresponding coke amounts by the mass of catalyst, which corresponds to the sample mass at the end of TGA procedure after the burning of hard coke. The total coke amount is the sum of coke precursors and hard coke amounts. The acidity of fresh catalyst and the free acidity of coked catalysts was measured by NH3 temperature-programmed desorption (TPD) with Micromeritics AutoChem 2910, using mild temperature pretreatment and combining of TPD without and with ammonia to avoid the falsification due to the chemically active character of coke precursors at high temperature during the TPD process.7 The same heating rate as that applied for TGA was applied to the TPD analysis. Around 50 mg coked catalyst was preheated with 10 K/min to 473 K for 60 min to remove water and reaction mixture components adsorbed on the catalyst as at the TGA analysis. The temperature then decreased to 353 K for ammonia (10% in helium) adsorption under a flow regime. After saturated with ammonia, the loaded sample was heated to 383 K with 10 K/min for physisorbed NH3 to be desorbed. The linear temperature program then started from 383 to 873 K with 10 K/min and remained at 873 K for 30 min. The desorbed ammonia and removed coke precursors were monitored continuously with a thermal conductivity detector (TCD). This is called first TPD. The fresh zeolite was also analyzed by this method to measure the total free acid sites. The same TPD was carried out initially without preceding

10.1021/ie071353a CCC: $40.75 © 2008 American Chemical Society Published on Web 03/22/2008

Ind. Eng. Chem. Res., Vol. 47, No. 9, 2008 2949

Figure 1. TGA and TPD temperature program.

ammonia adsorbed. This was called TPD without ammonia. Second TPD was carried out with 873 K preheating instead of 473 K. During this period of preheating, coke precursors as well as adsorbed water and reaction mixture components were removed. The first TPD contains both ammonia and coke precursors adsorbed on the zeolite. However, the TPD without ammonia only contains coke precursors. By subtracting the signal of TPD without ammonia from the signal of first TPD, the free acid sites of coked catalyst inhabited by both coke precursors and hard coke can be calculated. First TPD of fresh zeolite indicates the free acid sites of fresh zeolite. Because coke precursors have been removed after the pretreatment at 873 K in a helium stream during second TPD, only hard coke deposited on the catalyst. The signal of second TPD shows the free acid sites of coked zeolite inhabited only by hard coke. All of these TCD signals were normalized by the zeolite weight. With these methods, not only the concentration of acid sites but also the corresponding acid strength distribution could be determined. The TGA and TPD temperature programs are shown in Figure 1. 2.3. Catalyst Preparation. The USHY was calcined in an oven with 10 K/min heating rate to 873 K for 12 h. This is fresh catalyst for the reaction. Because the samples used were exposed to high-temperature cycles, to test their thermal stability the acidity of a regenerated catalyst was also measured. The NH3-TPD curve of the regenerated catalyst after removal of all coke components at 873 K in air was exactly the same as that of the fresh sample. The structure of the fresh catalyst samples being ultrastabilized zeolite and having been calcined at 873 K stays intact upon exposure at high temperatures. The selective poisoning of strong catalytic acid sites was carried out over fresh catalyst with 1-pentene reactions (P1-pentene ) 0.2 bar, PN2 ) 0.8 bar) in a fixed-bed reactor at different experimental conditions: (1) 573 K for 20 min of time-on-stream (TOS), and (2) 623 K for 300 min of TOS respectively. These two sets of conditions were selected from a wider set of experiments carried out because they produce a sample with almost no strong acidity present (2) and another one with almost all weak acidity intact (1) as shown below. After each reaction run, the coked catalyst was collected and thermally treated at the TGA equipment at 873 K (10K/min)

Table 1. Coke Content of PCS1 and PCS2 under the Specified Reaction Conditions samplesa

coke precursors %

hard coke %

total coke %

PCS1 (573 K, TOS ) 20 min) PCS2 (623 K, TOS ) 300 min)

6.7 (removed) 4.2 (removed)

11.9 17.7

18.6 21.9

a

% ) gcoke/100gzeolite anhydrous.

for 30 min in nitrogen flow to completely remove the coke precursors. We call these two catalyst samples produced at conditions 1 and 2 pre-coked sample 1 (PCS1) and pre-coked sample 2 (PCS2), respectively. Using the above-described coke classification method, the contents of the coke precursors and hard coke of PCS1 and PCS2 are shown in Table 1. The amount of total coke and hard coke is larger at a higher reaction temperature and longer TOS, whereas that of coke precursors is smaller.6 Furthermore, the change of total coke amount is not comparable to that of hard coke, which can be explained by previous work; fast transformation of coke precursors into hard coke compared to the much slower formation of reactant into coke precursors.6 Because coke precursors have been removed in PCS1 and PCS2, only hard coke remained deposited on these samples, which cannot be removed during the reaction experiments because the reaction temperature was 573 K, much below 873 K. Using the above-described TPD method (described in chapter 2.2), the free acid sites of fresh catalysts, PCS1 and PCS2, were determined and presented in part a of Figure 2.The acid sites of PCS2 were deactivated much more than those of PCS1. To further look inside strong and weak sites, the NH3-TPD thermogram of the fresh catalyst was deconvoluted using the digital deconvolution method of Micromeritics software. The correspondent deconvoluted curves of fresh catalyst are shown in part b of Figure 2. The original desorption curve starts at 380 K and ends at 780 K. Two deconvoluted peaks are located at 473 and 635 K, representing weak and strong-acid sites, respectively. The concentration of strong-acid sites is much lower than that of weak-acid sites. In the same figure, the weak and strong acidity curves of PCS1 and PCS2 are also shown. For PCS1 and PCS2, we made the reasonable assumption that the weak/strong-acid sites are a fraction of the weak/strong-acid sites of the fresh catalyst. We fitted the weak/strong sites’ fractions so that the sum of the curves of weak and strong sites shows the lowest deviation from

2950

Ind. Eng. Chem. Res., Vol. 47, No. 9, 2008

Figure 3. 1-pentene reaction network.

equipped with a flame ionization detector, using a 100 m nonpolar DB-Petro (J&W Scientific) capillary column (100 m × 0.25 mm × 0.5 µm). At the end of the experimental run, the saturator was bypassed and the reactor was disconnected from the rig and put in ice for quick cooling to ambient temperature. Coked or further coked catalyst samples were obtained at the end of each reaction and analyzed by the methods described in chapter 2.2. 3. Results and Discussion

Figure 2. (a) Acid sites distribution of fresh catalyst, PCS1 (pre-coked catalyst, deactivated at 573 K for 20 min with coke precursors removedonly hard coke remaining) and PCS2 (pre-coked catalyst, deactivated at 623 K for 300 min with coke precursors removed- only hard coke remaining). (b) Deconvolution into weak and strong-acid sites distribution of fresh catalyst, PCS1 (coked catalyst, deactivated at 573 K for 20 min with coke precursors removed- only hard coke remaining) and PCS2 (coked catalyst, deactivated at 623 K for 300 min with coke precursors removedonly hard coke remaining).

the actual TPD curve of the corresponding sample. About 60% of strong-acid sites remained in PCS1, whereas PCS2 has very few strong-acid sites (10%). The corresponding weak site fractions are 95% for PCS1 and 80% for PCS2. 2.4. Reaction Experiments. Catalytic reactions of 1-pentene over fresh catalyst, PCS1, and PCS2 were carried out at a temperature of 573 K and atmospheric pressure in a stainless steel tubular fixed-bed reactor. To ensure the same amount of pure USHY zeolite, 0.65 g of fresh catalyst, 0.73 g (0.65 g USHY + 0.08 g hard coke) of PCS1, and 0.77 g (0.65 g USHY + 0.12 g hard coke) of PCS2 were used in each experiment. The catalyst bed was placed between two metal meshes and steal wool to ensure isothermicity, and it was indeed checked to be isothermal using a thermocouple inserted in a small metal protection tube that was placed in the center of the reactor. In these experiments, the reactant partial pressure was P1-pentene ) 0.2 bar (PN2 ) 0.8 bar), the weight hour space velocity was WHSV ) 21.553 h-1, and the residence time was τ573K ) 0.057 s. Products were collected in a heated 10-way sampling valve at a specified TOS, namely 1, 2, 3, 5, 7, 9, 12, 15, and 20 min. Upon completion of the experiment, the samples were analyzed by a Hewlett-Packard 5890 packed series II gas chromatograph

3.1. Product Distribution and Conversion. The major products of 1-pentene reactions over different catalysts according to the type of reaction were (1) propene (C3d) and isobutene (iso-C4) produced by cracking (Cr), (2) n-pentane (n-C5) and 2-methylbutane (2-m-C4) produced by hydride transfer (HT), (3) 2-methyl-2-butene (2-m-2-C4d) and 2-methyl-1-butene (2-m-1-C4d) produced by skeletal isomerization (SkI), and (4) trans-2-pentene (trans-2-C5d), cis-2-pentene (cis-2-C5d ) produced by double-bond isomerization (DbI). A reaction network was suggested based on the product distribution as shown in Figure 3, supported also by results on other zeolite and zeotype acidic catalysts.8 Product profiles with TOS according to the reaction over all three catalyst samples are presented in Figures 4-7. The total amount of cracking products (Figure 4) over all three different catalysts decreases with TOS, more profoundly over fresh catalyst and PCS1, whose strong-acid sites were only partially poisoned to a relatively low degree. These products decreased drastically during the initial stage, indicating a fast deactivation of the cracking reaction. These phenomena concerning the decrease of cracking products can be explained by a rapid coke formation, which takes place on strong-acid sites, resulting in strong-acid site deactivation at the beginning of catalyst exposure to the reaction mixture.8,9 During the reaction of PCS2, whose strong-acid sites have been almost completely poisoned, only 2.7% of cracking products were produced at 1 min TOS, compared to 21.8% formed over fresh catalyst. Stronger-acid sites are expected to be more active for cracking proportionally to their strength.5 Figure 5 shows that hydrogen transfer was initially the predominant reaction, accounting for 56.6% over fresh catalyst and 47.6% over PCS1 at 1 min TOS, respectively. However, much less hydrogen transfer products were formed

Ind. Eng. Chem. Res., Vol. 47, No. 9, 2008 2951

Figure 4. Cracking products (C3d + iso-C4) of 1-pentene reaction over different catalysts at 573 K for 20 min.

Figure 5. Hydride transfer products (n-C5 + 2-m-C4) of 1-pentene reaction over different catalysts at 573 K for 20 min.

Figure 6. Double-bond isomerization products (trans-2-C5d + cis-2C5d) of 1-pentene reaction over different catalysts at 573 K for 20 min.

over PCS2 (4.5%) than the other two systems at 1 min. Furthermore, hydrogen transfer products show a similar pattern as cracking products. Then, they decrease drastically with TOS due to rapid coke formation at initial stage of the reaction.6,7,9 The composition of coke is aromatics10 whose carbon to

Figure 7. Skeletal isomerization products (2-m-1-C4d + 2-m-2-C4d) of 1-pentene reaction over different catalysts at 573 K for 20 min.

hydrogen ratio (C/H) is much larger than that of paraffins. Coke components are hydrogen poor with a carbon to hydrogen ratio (C/H) much larger than that of the reactant. During coking, hydrogen is transferred from coke to olefinic surface species, which desorb as paraffinic products. Formation of paraffins, n-pentane, 2-methyl-butane and isobutane, in these reactions, is enhanced by hydride transfer from these free hydrogens at initial TOS. As this hydrogen was consumed in conjunction with a sharp decrease in the coking rate, no more hydrogen was available for hydride transfer to form paraffins, resulting in a sharp drop of the yield of n-pentane, 2-methyl-butane, and isobutane from 1-pentene. For double-bond isomerization shown in Figure 6, the mole fraction of trans- and cis-2-pentene increases rapidly from less than 10% at 1 min TOS to more than 60% at 7 min, followed by a plateau at a considerably high level until 20 min over fresh catalyst and PCS1. From these profiles, it seems that trans- and cis-2-pentene isomers are intermediate products formed by 1-pentene and react further to cracking and hydride transfer reactions. The activities of cracking and hydride transfer decrease rapidly due to rapid coking of strong-acid sites, whereas isomerization maintains high activity, resulting in an increase in the mole fraction of 2-pentene isomers (trans- and cis-2-pentene). It can be deduced that the predominant reaction taking place after 7 min is double-bond isomerization due to the fast initial deactivation of strong-acid sites. A confirmation is provided by the profile over PCS2, where double-bond isomerization was the main reaction even at the beginning. Because almost no cracking or hydride transfer occurs over PCS2 due to the poisoning of strong-acid sites, trans- and cis-2-pentene do not react further. Furthermore, the level of double-bond isomers over PCS2 at 20 min remains higher than the corresponding value over PCS1 (Figure 6), which is in good agreement with the larger amount of coke components formed on PCS1 than on PCS2 (section 3.3, Table 2). As a result, the increase of their mole fractions takes place much earlier than over fresh and PCS1. For the same reason, the increase of trans- and cis-2-pentene takes place earlier over PCS1 than fresh due to availability originally of less strong sites which deactivate faster allowing double-bond isomerization to become the dominant reaction earlier. Figure 7 reveals a maximum in the skeletal isomerization products over fresh catalyst and PCS1, which indicates that skeletal isomerization products are also intermediates being formed by 1-pentene and undergoing further cracking/hydride transfer. The fact that the decline of the mole fractions of these products with TOS after

2952

Ind. Eng. Chem. Res., Vol. 47, No. 9, 2008

Table 2. Content of Coke Formed over Fresh Catalyst, PCS1, and PCS2 at 573 K and TOS ) 20 min (additionally Formed Hard Coke for PCS1 and PCS2) samplesa

coke precursors %

hard coke %

total coke %

fresh fresh with N2 purging PCS1 PCS2

6.7 6.6 5.1 3.3

11.9 11.7 4.7 2.8

18.6 18.3 9.8 6.1

a

% ) gcoke/100gzeolite anhydrous.

they reached their maximum is faster than the corresponding decline of double-bond isomerization products (Figure 6) means that strong-acid sites contribute a lot to the formation of these products. However, the fact that over PCS2 skeletal isomerization products have a higher mole fraction than cracking/hydride transfer products means that the acid strength needed for skeletal isomerization is not as high as the one needed for cracking/ hydride transfer. Generally, the acid strength required for these reactions decreases in the order: cracking ≈ hydride transfer > skeletal isomerization . double-bond isomerization.11 According to this, strong-acid sites will promote cracking and hydride transfer reactions, whereas weak-acid sites will be more selective toward skeletal isomerization. However, when the acidity is too low, the activity of the catalyst is only sufficient for double-bond isomerization.8 Moreover, because the order of strong-acid sites of these three catalysts is, fresh > PCS1 > PCS2 (almost no strong sites) at the beginning of the reaction, 1 min TOS, the selectivity of cracking products (fresh catalyst, 21.8%; PCS1, 16.0%; PCS2, 2.8%) and hydride transfer (fresh catalyst, 56.7%; PCS1, 47.5%; PCS2, 4.5%) decreases with decreasing concentration of strongacid sites, whereas double-bond isomerization (fresh catalyst, 3.1%; PCS1, 10.4%; PCS2, 64.2%) and skeletal isomerization (fresh catalyst, 2.6%; PCS1, 6.8%; PCS2, 12.5%) increase. The conversions of 1-pentene versus TOS are shown in part a of Figure 8 for the three different catalysts. As expected, the conversion eventually decreased at all catalysts because of catalyst deactivation. During the initial period, the reaction was accompanied by a deactivation phase, which was stronger over PCS1 compared to that of the fresh catalyst. Initially, the conversion was almost 100% over fresh catalyst and PCS1. As discussed above, during the first minutes of TOS the conversion was exclusively due to hydride transfer/cracking reactions and strong coking on strong-acid sites, whereas later it was due to isomerization reactions. To take into account the different deactivation behavior of various reaction groups, the conversion of all pentene linear isomers (1-pentene, cis-2-pentene, and trans-2-pentene) was also calculated. This conversion excludes double-bond isomerization. It accounts for all other reactions that undergo rapid deactivation, as reflected in the plot of the conversion of all of the pentene linear isomers (part b of Figure 8). Indeed, the conversion of all of the linear pentene isomers over PCS2 with almost no strong-acid sites is much lower than those over fresh catalyst and PCS1. The conversion of all of the linear pentene isomers over PCS1 with almost the same strength acidity as the fresh sample is practically the same as that over the fresh sample. 3.2. Purge with Nitrogen to Test Hydrogen Release Delay. Another question that arises regarding catalytic hydrocarbon reactions is one related to the initial fast deactivation in conjunction with the appearance of components produced by secondary reactions into the gas product spectrum. More specifically, the question tested is the following. Is the initial

Figure 8. (a) Conversion of 1-pentene over fresh catalyst, PCS1 (coked catalyst, deactivated at 573 K for 20 min with coke precursors removed, only hard coke remaining) and PCS2 (coked catalyst, deactivated at 623 K for 300 min with coke precursors removed, only hard coke remaining). (b) Conversion of all linear pentene isomers over fresh catalyst, PCS1 (coked catalyst, deactivated at 573 K for 20 min with coke precursors removed, only hard coke remaining) and PCS2 (coked catalyst, deactivated at 623 K for 300 min with coke precursors removed, only hard coke remaining).

deactivation extremely rapid (almost instantaneous) with the result of an immediate complete decline of cracking/hydride transfer reactions? In this scenario, hydride transfer products would be belatedly released because of hydrogen release delay. The alternative scenario would be that the fast initial deactivation is not extremely rapid and the decline of the cracking/hydride transfer products simply follows the catalyst deactivation. To clarify this question, a reaction of 1-pentene over fresh USHY was carried out at the same temperature of 573 K. The reaction conditions and procedure were the same as described in the experimental section with the following modification. After collecting the first sample, the feeding of reactant 1-pentene stopped at TOS ) 1.5 min, and the fixed-bed reactor was purged with pure nitrogen for 2 min. The choice of the purge timing was justified as follows. Because the coke formation rate was at its highest before 1 min TOS,6 the nitrogen purge was carried out after the first sample was collected at 1 min TOS. From previous work with TGA and TPD analysis,6,12 there is no profound change in coke content and character if the coked catalyst is purged in nitrogen at 573 K for only 2 min. After nitrogen purging finished (original TOS ) 3.5 min), reactant continued to be fed into the bed and the reactor started to operate in reaction mode again. Reaction mixture sampling

Ind. Eng. Chem. Res., Vol. 47, No. 9, 2008 2953

Figure 9. Major product distribution vs TOS during 1-pentene reactions over fresh USHY catalyst at 573 K with and without purging N2.

Figure 10. TPD without ammonia of deactivated USHY zeolite coked over different catalysts (T ) 573 K, TOS ) 20 min).

continued as usual. The results of product distribution during this experiment were compared with the product distribution of the original experiment. For comparison reasons, we would like to distinguish between original TOS and modified TOS. Original TOS is the experimental time counting from the original start of the experiment when reactant was fed into the catalyst bed for the first time. Until 1.5 min, the modified TOS is the same as the original one. After 3.5 min, that is, after finish of nitrogen purging, modified TOS is equal to the original one reduced by 2 min, that is, the time period of the nitrogen purging. The sampling during this experiment took place at: Modified TOS: 1, 2, 3, 5, 7, 9, 12, 15, 20 min Original TOS: 1, 4, 5, 7, 9, 11, 14, 17, 22 min The product distribution of this experiment at the modified TOS should compare with that of the original experiment at the original TOS. If the product distribution profiles are almost the same, then there is no hydrogen release delay taking place. If not, then hydrogen release delay distorts the picture of product distribution. The product distributions of 1-pentene reaction (four major products) over fresh USHY zeolite without and with nitrogen purging are presented in Figure 9. It can be seen that there are no significant differences in distribution of all of the products after purging with nitrogen. This means that the decline of the yield of cracking/hydride transfer products follows the initial strong decrease of the catalyst activity rather than being released belatedly due to hydrogen release delay. Furthermore, we have explained the effect of the nitrogen purge on the coke component concentration. After 20 min of the reaction, the coked catalyst was analyzed by TGA, and the result is displayed in Table 2. The concentrations of coke precursors, hard coke, and total coke formed over USHY zeolite during the N2 purge experiment are practically identical to those formed over USHY zeolite during a standard experiment. N2 purging had no effect on coke formation either. 3.3. Coke Formation and Acid Site Characterization. Table 2 shows also the content of coke precursors, hard coke, and total coke over fresh catalyst, PCS1, and PCS2 after their deactivation. We would like to clarify that for the pre-coked samples, PCS1 and PCS2, these coke amounts refer to additional coke components formed during the respective experiments, and they do not include the hard coke that was already formed during catalyst preparation. Both coke precursor and hard coke

concentrations decrease with decreasing catalyst acidity. Strongacid sites favor cracking and hydride transfer reactions as well as coking.5,6 The coke precursor distribution taken from a TPD without NH3 can be seen in Figure 10, where the peaks from the TCD output are due to coke precursors in the carrier gas. Because the TCD signal response factors of different coke precursors are not profoundly different,7 the integral TPD area of the different catalyst systems is in the same order as the coke precursors content measured by TGA. Moreover, the TPD signal of coke precursors formed on fresh catalyst shows a wide distribution, whereas coke precursors formed on PCS1 and PCS2 locate within that of fresh catalyst. In previous work,6,7 we further classified coke precursors into large/stable coke precursors, showing a peak at high temperature, and small/unstable coke precursors, showing a peak at low temperatures. From Figure 10, we can see there are two peaks located at 650 and 750 K correspondingly in TCD curves of fresh catalyst and PCS1, which are due to small/unstable coke precursors and large/stable coke precursors, respectively. Whereas the peaks corresponding to small/unstable coke precursors have declined relatively little at PCS1 and PCS2, the decline of the large/ stable coke precursor’s peak is profound for both samples (considerably more for PCS2 than for PCS1), indicating that lack of strong acidity slows down coke growth much more than coke precursor formation. Additionally, hard coke formed over PCS1 is lower than that over fresh catalyst and even lower than that over PCS2. We are going to employ the assistance of a model of acid site deactivation shown in Figure 11 to explain this. In the model, all of the strong-acid sites of PCS2 have been blocked as well as some of its weak sites, whereas for PCS1 only a small part of strong and weak sites has been blocked, to approach the picture of free strong- and weak-acid sites for all of the catalysts (part b of Figure 2). There is only one peak in the PCS2 curve (670 K) arising from small/unstable coke precursors formed on weak-acid sites. Furthermore, because there are more free acid sites in fresh catalyst than in PCS1, more large/stable coke precursors are formed over fresh catalyst than over PCS1. Hence, more coke precursors are removed at high TPD temperature from fresh catalyst than from PCS1. This agrees well with coke being formed preferentially on the strongest acid sites.7,12 Parts a-c of Figure 12 present the original available total acid sites, second TPD (free acid sites not blocked by hard coke),

2954

Ind. Eng. Chem. Res., Vol. 47, No. 9, 2008

Figure 11. Acid sites deactivation model.

and free acid sites of the three catalyst systems, respectively. The gap area between total acid sites and second TPD is proportional to the concentration of acid sites inhabited by the additionally formed hard coke. The gap area between second TPD and free acid sites is the concentration of acid sites occupied by coke precursors. For all three systems, the integral area of acid sites due to hard coke poisoning is less than that of coke precursors. However, from the TGA results in Table 2, the content of hard coke is larger than that of coke precursors over fresh catalyst. If we set

area of total number of acid sites occupied by hard coke area of 2nd TPD ∝ R) weight content ghard coke/100gzeolite of hard coke area of 2nd TPD number of acid sites area of free occupied by coke precursors ∝ β) weight content of gcoke precursors/100gzeolite coke precursors then: Rfresh(0.79), RPCS2(5.36), RPCS1(5.74), βfresh(9.66), βPCS1(8.22), and βPCS2(10.58) The R values indicate hard coke species, which are formed over very strong sites on fresh catalyst (coordinates (5,6) and (6,6) in Figure 11) and are much larger/heavier than hard coke species formed over precoked samples, PCS1 and PCS2, and even much larger and heavier than coke precursors formed on weak-acid sites. From β values, we can conclude that the acid sites occupied by coke precursors are comparable to the content of coke precursors, which suggests that coke precursors deposited on weaker acid sites of each catalyst are relatively uniform. The difference may come from the different types of coke precursors, which may not have a significantly different molecular weight. 4. Conclusions

Figure 12. (a) TPD of total acid sites, free acid sites, and second TPD of deactivated fresh catalyst coked during 1-pentene reactions at 573 K and TOS ) 20 min. (b) TPD of total acid sites, free acid sites, and second TPD of deactivated PCS1 coked during 1-pentene reactions at 573 K and TOS ) 20 min. (c) TPD of total acid sites, free acid sites, and second TPD of deactivated PCS2 coked during 1-pentene reactions at 573 K and TOS ) 20 min.

USHY zeolites whose strong-acid sites were selectively poisoned by hard coke were tested in a 1-pentene reaction, and their performance was compared with fresh USHY zeolite to study the role of strong-acid sites. It appears that cracking and hydride transfer is catalyzed by strong-acid sites, whereas weakacid sites catalyze only double-bond isomerization. The hydride

transfer products are formed on strong-acid sites directly instead of hydrogen release delay. On the other hand, strong-acid sites play a significant role in coke formation. Moreover, coke formed on strong-acid sites is much heavier than that on weak-acid sites. The novel TPD method, mild temperature pretreatment combined with TPD without, and with ammonia can provide not

Ind. Eng. Chem. Res., Vol. 47, No. 9, 2008 2955

only the amount of acid sites but also their strength distribution. The amount of coke precursors and their removal ability can be obtained from TPD without NH3, and this can be used to further study the formation of coke precursors on acid sites. Acknowledgment Baodong Wang would like to acknowledge financial support of Overseas Research Students Awards Scheme and K. C. Wong Scholarship. The authors would like to thank Steve Coulson, Micromeritics, for his help with deconvolution of the fresh catalyst TPD curve. Literature Cited (1) Borges, P.; Pinto, R.; Lemos, M.; Lemos, F. Activity-acidity relationship for alkane cracking over zeolites: n-hexane cracking over HZSM-5. J. Mol. Catal. A: Chem. 2005, 229, 127. (2) Gates, B. C. Catalytic Chemistry; Wiley: New York, 1992. (3) Guisnet, M.; Magnoux, P. Organic chemistry of coke formation. Appl. Catal., A 2001, 212, 83. (4) Chen, S.; Manos, G. Study of coke and coke precursors during catalytic cracking of n-hexane and 1-hexene over ultrastable Y zeolite. Catal. Lett. 2004, 96, 195.

(5) Williams, B. A.; Babitz, S. M.; Miller, J. T.; Snurr, R. Q.; Kung, H. H. The roles of acid strength and pore diffusion in the enhanced cracking activity of steamed Y zeolites. Appl. Catal., A 1999, 177, 161. (6) Wang, B.; Manos, G. A novel thermogravimetric method for coke precursor characterisation. J. Catal. 2007, 250, 121. (7) Wang, B.; Manos, G. Acid site characterisation of coked USHY zeolite using Temperature Programmed Desorption with a component nonspecific detector. Ind. Eng. Chen. Res. 2007, 46, 7977. (8) Hochtl, M.; Jentys, A.; Vinek, H. Isomerization of 1-pentene over SAPO, CoAPO (AEL, AFI) molecular sieves and HZSM-5. Appl. Catal., A 2001, 207, 397. (9) Brillis, A.; Manos, G. Ind. Eng. Chen. Res. 2003, 42, 2292. (10) Henriques, C. A.; Magnoux, P.; Guisnet, M. Characterization of the Coke Formed Duringo-Xylene Isomerization over Mordenites at Various Temperatures. J. Catal. 1997, 172, 436. (11) Corma, A.; Wojciechowski, B. W. The Catalytic Cracking of Cumene. Catal. ReV.sSci. Eng. 1982, 24, 1. (12) Guisnet, M.; Magnoux, P. Fundamental description of deactivation and regeneration of acid zeolites. Stud. Surf. Sci. Catal. 1994, 88, 53.

ReceiVed for reView October 8, 2007 ReVised manuscript receiVed February 15, 2008 Accepted February 19, 2008 IE071353A